Estimates, power and sample size calculations for two-sample ordinal outcomes under before-after study designs

Area of Science:

• Biostatistics
• Clinical Trial Design
• Ordinal Data Analysis

Background:

• Comparing intervention effects on ordinal outcomes requires robust statistical methods.
• Pre-post intervention designs are common in clinical research.
• Accurate sample size determination is crucial for study power.

Purpose of the Study:

• To develop and present sample size calculations for comparing two groups with ordinal outcomes in pre-post intervention studies.
• To provide flexible methods applicable to various ordinal response models.
• To evaluate the performance of the proposed sample size calculations.

Main Methods:

• The study utilizes log-odds models incorporating random intercepts, treatment, time, and treatment-by-time interaction terms.
• Calculations accommodate flexible odds ratio assumptions, including proportional odds and adjacent category models.
• Simulation studies are employed to assess the accuracy of the sample size calculations.

Main Results:

• The proposed sample size calculations are presented for comparing ordinal outcomes between intervention groups.
• The methods are flexible, accommodating various models for ordinal data.
• Simulation results indicate that the nominal and actual powers of the test are similar for given sample sizes.

Conclusions:

• The developed sample size calculations are suitable for pre-post intervention studies with ordinal outcomes.
• The flexible modeling approach enhances the applicability of these calculations.
• The findings support the use of these methods for ensuring adequate statistical power in clinical trials.