Comment and Completion: Implementation of Parallelism Testing for Four-Parameter Logistic Model in Bioassays.

Parallelism testing between two four-parameter logistic curves has been widely discussed over the last decade. Current tests available in common statistical software used in laboratories have been shown to be highly flawed. In 2012, Yang et al. showed an easy way to implement an intersection union test based on confidence intervals on ratios of parameters of both curves. The method was automated using a fully good manufacturing practice-compliant software package. Although the rationale is correct and efficient, a small mistake appears in the computation of the confidence intervals in the paper and may lead to error when implementing the intersection union test in a software package. Because parallelism testing is both a prerequisite for the determination of relative potency of bioassays and a regulatory requirement, it is important to rectify this mistake. In this paper, we show the actual formulas to be used to compute confidence interval on ratios of parameters.