Response to criticisms of the US FDA parametric approach for withdrawal time estimation: rebuttal and comparison to the nonparametric method proposed by Concordet and Toutain.

The benefits and drawbacks of using nonparametric methods for estimating product withdrawal times have been debated for many years. This issue was recently revived by Concordet & Toutain (1997a, b) when they described a nonparametric method for withdrawal time estimation. The authors urged the international adoption of this approach, basing their recommendation on three fundamental concerns: (1) the lack of a consistent official procedure for determining a withdrawal time within the European Union (EU); (2) the need to identify a statistical method for improving the international harmonization of withdrawal times for new chemical entities; and (3) a lack of confidence in the robustness of the US Food and Drug Administration/Center for Veterinary Medicine (US FDA) procedure, particularly with respect to minor violations in the underlying parametric assumptions. Due to the critical nature of these issues, the US FDA considers it vital to respond to these concerns. This paper provides a description of the US FDA parametric procedure. We also examine the statistical concerns expressed by Concordet and Toutain, identifying the reasons for our confidence in the US FDA parametric approach. Finally, using their Monte Carlo simulation models, we generate additional datasets to explore the behaviour of their nonparametric procedure and evaluate its ability to support US FDA regulatory activities.