Comparison of different methods for decision-making in bioequivalence assessment

Area of Science:

• Biostatistics
• Pharmacometrics
• Regulatory Science

Background:

• Bioequivalence assessment is crucial for drug approval.
• Traditional methods often rely on symmetrical confidence intervals.
• Regulatory requirements can be either symmetrical or asymmetrical.

Purpose of the Study:

• To compare the performance of symmetrical confidence intervals versus shortest confidence intervals for bioequivalence assessment under different regulatory conditions.
• To provide recommendations for the optimal use of statistical methods in bioequivalence studies.

Main Methods:

• Simulations were conducted to evaluate the properties of different confidence interval approaches.
• Performance was assessed based on level properties and statistical power.
• Analysis considered both symmetrical and asymmetrical regulatory scenarios.

Main Results:

• Under symmetrical regulatory requirements, shortest confidence intervals demonstrated superior power compared to symmetrical confidence intervals.
• Symmetrical confidence intervals are not recommended for asymmetrical regulatory requirements.
• Posterior probabilities offer a viable alternative for decision-making in asymmetrical cases.

Conclusions:

• The choice of confidence interval method for bioequivalence should align with the nature of regulatory requirements.
• Symmetrical confidence intervals alone can be misleading; consider posterior probabilities or shortest confidence intervals for comprehensive inference and interpretation.
• A loss function approach is necessary for choosing between methods when regulatory requirements are symmetrical.