Assessing individual equivalence in parallel group and crossover designs: Exact test and sample size procedures

Area of Science:

• Biostatistics
• Clinical Trial Design
• Statistical Inference

Background:

• Individual equivalence testing is crucial for comparing treatments, offering an alternative to average equivalence.
• Current methods, such as extensions of two one-sided tests (TOST), often lack proper Type I error control and are overly conservative.
• Existing tolerance interval approaches for individual equivalence may not be statistically rigorous.

Purpose of the Study:

• To develop and present exact statistical tests for assessing individual treatment equivalence.
• To address the limitations of existing TOST-based methods for individual equivalence.
• To provide a more accurate and powerful approach for comparative studies.

Main Methods:

• Development of exact tests for individual equivalence under parallel group and crossover designs.
• Rigorous evaluation comparing critical values and Type I error probabilities of new and existing methods.
• Simulation studies to justify power and sample size calculations for the proposed approach.

Main Results:

• The proposed exact tests demonstrate superior control over Type I error compared to TOST extensions.
• Evaluations reveal significant discrepancies in critical values and error rates, highlighting TOST shortcomings.
• The new methodology offers improved statistical power for detecting individual equivalence.

Conclusions:

• The presented exact tests provide a statistically sound and advantageous method for assessing individual treatment equivalence.
• Existing TOST-based procedures for individual equivalence have critical limitations that are resolved by the proposed approach.
• This work offers a more reliable framework for clinical trial analysis and treatment comparisons.