Optimal and/or Efficient Two treatment Crossover Designs for Five Carryover Models.

Area of Science:

• Clinical Trials
• Biostatistics
• Experimental Design

Background:

• Crossover designs are crucial in clinical trials when carryover effects are uncertain.
• Existing designs are optimized for specific carryover models, often neglecting subject numbers.
• Subject numbers significantly influence design optimality and practical implementation.

Purpose of the Study:

• To develop a robust method for identifying optimal two-treatment crossover designs.
• To account for subject numbers in design selection for clinical trials.
• To evaluate designs across five different carryover models, including newly proposed ones.

Main Methods:

• A 5M algorithm was developed and implemented in a computer code.
• The algorithm systematically searched all possible two-treatment crossover designs.
• Designs were evaluated for optimality and efficiency under five distinct carryover models.

Main Results:

• The algorithm identified optimal and/or efficient crossover designs for two, three, and four periods with 4 to 20 subjects.
• Twenty-four new optimal designs were found for established carryover models.
• Thirty-four designs demonstrated optimality for the newly added carryover models.

Conclusions:

• The 5M algorithm provides a comprehensive approach to selecting robust crossover designs.
• The findings offer a valuable resource for researchers designing clinical trials with unknown carryover effects.
• Consideration of subject numbers leads to more practical and efficient trial designs.