Helmut Schütz1. 1. BEBAC, Neubaugasse 36/11, 1070, Vienna, Austria, helmut.schuetz@bebac.at.
Abstract
PURPOSE: The aim of this study is to assess the current status of non-fixed sample size designs in bioequivalence trials with a focus on two-stage adaptive approaches. METHODS: We searched PubMed and Google Scholar from inception to October 2014. Regulatory guidelines were obtained from the public domain. Different methods were compared by Monte Carlo simulations for their impact on the patient's and producer's risks. RESULTS: Add-on designs, group sequential designs and adaptive two-stage sequential designs are currently accepted to demonstrate bioequivalence in various regulations. All three approaches may inflate the patient's risk if applied inconsiderately. Direct transfer of methods developed for superiority testing to bioequivalence is not warranted. Published two-stage frameworks maintain the type I error and generally the desired power. Adaptation based on the observed T/R ratio observed in the first stage should be applied with caution. Monte Carlo simulations are an efficient tool to explore the operating characteristics of methods. CONCLUSIONS: Validated two-stage frameworks can be applied without requiring the sponsor to perform own simulations-which could further improve power based on additional assumptions. Two-stage designs are both ethical and economical alternatives to fixed sample designs.
PURPOSE: The aim of this study is to assess the current status of non-fixed sample size designs in bioequivalence trials with a focus on two-stage adaptive approaches. METHODS: We searched PubMed and Google Scholar from inception to October 2014. Regulatory guidelines were obtained from the public domain. Different methods were compared by Monte Carlo simulations for their impact on the patient's and producer's risks. RESULTS: Add-on designs, group sequential designs and adaptive two-stage sequential designs are currently accepted to demonstrate bioequivalence in various regulations. All three approaches may inflate the patient's risk if applied inconsiderately. Direct transfer of methods developed for superiority testing to bioequivalence is not warranted. Published two-stage frameworks maintain the type I error and generally the desired power. Adaptation based on the observed T/R ratio observed in the first stage should be applied with caution. Monte Carlo simulations are an efficient tool to explore the operating characteristics of methods. CONCLUSIONS: Validated two-stage frameworks can be applied without requiring the sponsor to perform own simulations-which could further improve power based on additional assumptions. Two-stage designs are both ethical and economical alternatives to fixed sample designs.