Randomization tests for assessing the equality of area under curves for studies using destructive sampling

Area of Science:

• Pharmacokinetics and Pharmacometrics
• Biostatistics
• Drug Development

Background:

• Testing area under the curve (AUC) equality is crucial for assessing bioequivalence in pharmacokinetic studies.
• Current methods often rely on normal distribution assumptions, which may not hold for small sample sizes or non-normal response distributions.
• Estimating AUC using linear combinations of response means is common when experimental units are sacrificed.

Purpose of the Study:

• To evaluate the applicability of normal-theory statistical tests for AUC equality in pharmacokinetic research.
• To propose and discuss randomization tests as a robust alternative for non-normal data and small sample sizes.
• To address the assessment of dose proportionality alongside AUC equality testing.

Main Methods:

• Utilized standard statistical theory to derive the distribution of AUC estimators based on response means.
• Investigated the limitations of normal-theory tests when response distributions are non-normal and sample sizes are small.
• Proposed and discussed the application of randomization tests as an alternative statistical approach.

Main Results:

• Normal-theory tests for AUC equality may be questionable with non-normal response distributions and small sample sizes.
• Randomization tests are presented as a valuable and applicable alternative to normal-theory tests in such scenarios.
• The study also touches upon methods for assessing dose proportionality.

Conclusions:

• Randomization tests offer a more reliable approach for testing AUC equality in pharmacokinetic studies with non-normal data and limited observations.
• These tests provide a robust alternative to traditional methods, enhancing the validity of bioequivalence assessments.
• Further consideration of dose proportionality is recommended in conjunction with AUC analysis.