Estimation of direct formulation effect under log-normal distribution in bioavailability/bioequivalence studies.

Area of Science:

• Pharmacokinetics and Biopharmaceutics
• Statistical Modeling in Clinical Trials

Background:

• Accurate estimation of relative bioavailability is crucial for bioequivalence studies.
• Log-normal distribution is often assumed for pharmacokinetic data in 2x2 crossover studies.
• Existing estimators may exhibit bias, impacting bioequivalence conclusions.

Purpose of the Study:

• To propose and evaluate an improved estimator for relative bioavailability on the original scale.
• To compare the performance of different estimators, including the proposed one, in 2x2 crossover bioequivalence studies.
• To provide a statistically robust method for bioequivalence assessment.

Main Methods:

• Development of the minimum variance unbiased estimator (MVUE) for relative bioavailability.
• Derivation of the bias and mean square error for various estimators.
• Simulation study to compare theoretical and empirical performance of estimators.
• Application of a numerical example to demonstrate the proposed estimation procedure.

Main Results:

• The proposed minimum variance unbiased estimator (MVUE) is identified as the maximum likelihood estimator adjusted for bias.
• Theoretical and simulation results indicate that the MVUE consistently outperforms other methods in terms of bias and mean square error.
• The MVUE is shown to be unbiased and provides the minimum variance among unbiased estimators.

Conclusions:

• The minimum variance unbiased estimator (MVUE) is the preferred method for estimating relative bioavailability in log-normal data from 2x2 crossover studies.
• Adoption of the MVUE can lead to more reliable bioequivalence decisions.
• The proposed method offers a statistically sound approach for pharmacokinetic data analysis.