A parametric confidence interval for a moment-based scaled criterion for individual bioequivalence

Area of Science:

• Pharmacokinetics and Biopharmaceutics
• Statistical Modeling in Drug Development
• Regulatory Science

Background:

• Determining exact confidence intervals for individual bioequivalence metrics is challenging due to unknown stochastic distributions.
• Existing methods for individual bioequivalence lack precise distributional properties, complicating regulatory assessment.

Purpose of the Study:

• To derive an exact distribution for a modified Relative Individual Risks (RIR) moment-based scaled statistic.
• To establish a reliable method for calculating confidence intervals for individual bioequivalence.
• To evaluate the performance of the proposed metric against existing methods.

Main Methods:

• Modified the Relative Individual Risks (RIR) moment-based scaled statistic to achieve an exact noncentral Fisher's F distribution.
• Approximated the noncentral F distribution with a central F distribution with adjusted degrees of freedom.
• Simulated bioequivalence studies using two formulations in a fully replicated design to compare performance.

Main Results:

• The modified RIR statistic follows an exact noncentral Fisher's F distribution, allowing for closed-form confidence bounds.
• The proposed confidence interval method was found to be slightly less biased and more precise than the unweighted metric.
• The performance was evaluated through simulations comparing experimental bioequivalence rates.

Conclusions:

• The modified RIR metric provides a statistically sound basis for individual bioequivalence assessment.
• This approach simplifies confidence interval determination, enhancing the reliability of bioequivalence testing.
• The findings support the use of this improved metric in regulatory decision-making for drug formulations.