Sample size calculations for group randomized trials with unequal group sizes through Monte Carlo simulations

Area of Science:

• Biostatistics
• Public Health Research
• Clinical Trial Design

Background:

• Group randomized trials are prevalent in cancer prevention and health promotion.
• Standard sample size calculations often assume equal group sizes, which is frequently not met in practice, leading to reduced statistical power.
• Existing methods for unequal group sizes are often limited to continuous outcomes.

Purpose of the Study:

• To develop a flexible method for sample size calculation in group randomized trials with unequal group sizes.
• To address limitations of current methods by accommodating various outcome types and study designs.

Main Methods:

• Development of a sample size calculation method using Monte Carlo simulation within a mixed-effects model framework.
• The method explicitly incorporates the variation in group sizes.
• Applicable to different outcome types (continuous and binary) and common group randomized trial designs (e.g., pre-and-post, cohort).

Main Results:

• The proposed Monte Carlo simulation approach effectively handles unequal group sizes in sample size calculations for group randomized trials.
• Demonstrated applicability to both continuous and binary outcomes through simulations.
• Validated through analysis of a real-world group randomized trial dataset.

Conclusions:

• The developed method provides a robust and adaptable tool for sample size determination in group randomized trials with unequal group sizes.
• This approach can improve the statistical power and reliability of findings in health promotion and cancer prevention research.
• The method is practical and can be readily implemented for various study designs.