Robust and realistic approaches to carry-over

Area of Science:

• Clinical Trials Methodology
• Biostatistics
• Experimental Design

Background:

• Crossover designs are widely used in clinical research to compare treatments within subjects.
• Carry-over effects, where a previous treatment influences the response to a subsequent one, can bias results.
• Statistical models are essential for analyzing crossover trial data and accounting for carry-over.

Purpose of the Study:

• To investigate the interplay between the choice of statistical model for carry-over effects and the selection of efficient crossover designs.
• To evaluate the impact of statistical model miss-specification on the efficiency of crossover designs.
• To re-examine previous claims regarding the efficiency of specific crossover designs.

Main Methods:

• Analysis focused on two-treatment, four-period crossover designs with two sequences.
• The study examined the consequences of miss-specifying the statistical model used to analyze carry-over effects.
• Simulations or theoretical derivations were likely employed to assess design efficiency under various model assumptions.

Main Results:

• The relationship between the carry-over model and the efficiency of crossover designs is complex.
• Model miss-specification can significantly impact the estimated efficiency of crossover designs.
• Previously reported claims of high efficiency for certain designs may be questionable when model miss-specification is considered.

Conclusions:

• The choice of statistical model for carry-over effects is critically important for valid crossover trial analysis.
• Researchers should carefully consider potential model miss-specification when selecting and interpreting crossover designs.
• Claims of design efficiency require robust validation against potential model miss-specification.