How to analyze continuous and discrete repeated measures in small-sample cross-over trials?

Area of Science:

• Biostatistics
• Clinical Trials
• Rare Diseases

Background:

• Repeated measures cross-over designs offer advantages for small-sample rare disease trials.
• Analyzing clustered within-subject and within-treatment period data in these trials presents challenges.

Purpose of the Study:

• To compare statistical methodologies for analyzing repeated measures cross-over designs in small-sample rare disease trials.
• To provide recommendations for optimal data analysis based on outcome type and treatment effect duration.

Main Methods:

• A real-data simulation study based on an epidermolysis bullosa simplex trial.
• Comparison of non-parametric marginal models, generalized pairwise comparison models, GEE-type models, and parametric model averaging.
• Analysis of both repeated binary and count data.

Main Results:

• The choice of methodology depends on the outcome type and the number of time points affected by treatment.
• Non-parametric marginal models are suitable for detecting differences in longitudinal profile shapes.
• Parametric model averaging is recommended for binary outcomes with single time point treatment effects; generalized pairwise comparisons are recommended otherwise.

Conclusions:

• Specific statistical methods are recommended for analyzing rare disease trials with repeated measures cross-over designs.
• Generalized pairwise comparison and parametric model averaging offer interpretable effect sizes and handle incomplete data effectively.