Power and sample size calculations for evaluating mediation effects in longitudinal studies.

Area of Science:

• Biostatistics
• Longitudinal Data Analysis
• Mediation Analysis

Background:

• Current power and sample size calculations for longitudinal mediation studies lack closed-form formulas.
• Missing data, common in longitudinal designs, pose a significant challenge.

Purpose of the Study:

• To develop and evaluate methods for power and sample size calculations in longitudinal mediation analysis.
• To address the challenge of missing data in these calculations.

Main Methods:

• Utilized the product of coefficients to measure longitudinal mediation effects.
• Evaluated three hypothesis testing methods: joint significance, normal approximation, and the test of b.
• Derived formulas for power and sample size calculations, incorporating missing data considerations.

Main Results:

• Provided formulae for power and sample size calculations under three distinct testing methods.
• Examined the performance of these methods with limited sample sizes via simulation.
• Demonstrated the application of the methods using an example from the Einstein aging study.

Conclusions:

• The study offers practical tools for designing longitudinal mediation studies.
• The developed methods effectively account for missing data, enhancing study design robustness.
• The findings are applicable to various fields employing longitudinal mediation analysis.