Abstract
The objective of average (bio)equivalence tests is to determine whether a parameter, such as the mean variation in treatment response between two conditions, lies within a specified equivalence range, indicating that the means of the conditions are equivalent. The widely-used Two One-Sided Tests (TOST) procedure checks if the target parameter is significantly greater or lower than pre-defined upper and lower equivalence limits by examining whether its confidence interval falls within these limits. However, the TOST procedure can be overly conservative and may quickly lose power for highly variable responses, in many cases reaching a flat zero over the entire parameter space, resulting in its inability to conclude equivalence when it truly exists. To address this limitation, we propose a new procedure called the α-TOST that incorporates a finite sample and variability correction by adjusting the size of the TOST to ensure a type I error rate of α in all situations. Our analysis shows that the α-TOST is uniformly more powerful, simple to compute, and outperforms its competitors in terms of operating characteristics. We use a case study of econazole nitrate deposition in porcine skin to illustrate the advantages of our approach over other available procedures.
Competing Interest Statement
The authors have declared no competing interest.
Footnotes
Correction of typos and ambiguous sentences only. Overall this revised version increases clarity and readability.
