Generalization of distribution--free confidence intervals for bioavailability ratios

Area of Science:

• Pharmacokinetics
• Biostatistics
• Drug Development

Background:

• Bioavailability assessment traditionally relies on confidence intervals, often using a 95% confidence level.
• Assessing the probability of bioavailability falling within specific critical values or bioequivalence ranges is of practical interest.
• Existing posterior probability distributions are often based on Analysis of Variance (ANOVA), which assumes a normal distribution.

Purpose of the Study:

• To generalize a distribution-free confidence interval method for bioavailability assessment.
• To enable the calculation of confidence probabilities for any given confidence limits.
• To provide a more robust alternative to ANOVA-based methods, particularly for non-normally distributed data.

Main Methods:

• Generalization of a distribution-free confidence interval based on the Wilcoxon signed-rank statistic.
• Application of the method to obtain confidence probabilities for specified bioavailability limits.
• Comparison of results with ANOVA-based posterior probability distributions.

Main Results:

• The generalized distribution-free method yields confidence probabilities for any desired confidence limits.
• Results closely align with ANOVA-based methods for unimodal and symmetrical distributions.
• The distribution-free approach provides more accurate and valid information for skewed or multimodal sampling distributions.

Conclusions:

• The proposed distribution-free method offers a valuable alternative for bioavailability and bioequivalence assessment.
• It overcomes the restrictive assumptions of traditional ANOVA-based methods.
• This approach enhances the reliability of bioavailability probability estimations, especially in complex data scenarios.