Reference datasets for bioequivalence trials in a two-group parallel design

On implementing Jeffreys' substitution likelihood for Bayesian inference concerning the medians of unknown distributions

When statisticians are uncertain as to which parametric statistical model to use to analyse experimental data, they will often resort to a non-parametric approach. The purpose of this paper is to provide insight into a simple approach to take when it is unclear as to the appropriate parametric model and plan to conduct a Bayesian analysis. I introduce an approximate, or substitution likelihood, first proposed by Harold Jeffreys in 1939 and show how to implement the approach combined with both a non-informative and an informative prior to provide a random sample from the posterior distribution of the median of the unknown distribution. The first example I use to demonstrate the approach is a within-patient bioequivalence design and then show how to extend the approach to a parallel group design.

Reference Datasets for Studies in a Replicate Design Intended for Average Bioequivalence with Expanding Limits

In order to help companies qualify and validate the software used to evaluate bioequivalence trials in a replicate design intended for average bioequivalence with expanding limits, this work aims to define datasets with known results. This paper releases 30 reference datasets into the public domain along with proposed consensus results. A proposal is made for results that should be used as validation targets. The datasets were evaluated by seven different software packages according to methods proposed by the European Medicines Agency. For the estimation of CV_wR and Method A, all software packages produced results that are in agreement across all datasets. Due to different approximations of the degrees of freedom, slight differences were observed in two software packages for Method B in highly incomplete datasets. All software packages were suitable for the estimation of CV_wR and Method A. For Method B, different methods for approximating the denominator degrees of freedom could lead to slight differences, which eventually could lead to contrary decisions in very rare borderline cases.

On the Incorrect Statistical Calculations of the Kinetica Software Package in Imbalanced Designs

This regulatory note supports the previous findings that suggest that the software package Kinetica, up to version 5.0.10, provides incorrect results for the 90% confidence intervals for the ratio test/reference where the groups are imbalanced in 2 × 2 crossover designs and parallel designs. The incorrect calculation results from using the simplified formula that is shown as an example in the Canadian guideline for a balanced dataset, but which provides an erroneous point estimate and confidence interval width in cases of imbalanced designs. Importantly, this software is rarely used for regulatory submissions in the European Union according to the search conducted in the Spanish Agency for Medicines and Health Care Products. According to our data, the error is minor if the imbalance between groups is small. However, the error may be relevant if the sample size is small and the imbalance is large. Therefore, bioequivalence studies should be reanalyzed by regulatory agencies to confirm the submitted results.