All Times EDT
Keywords: sample size calculation; confidence interval estimation; non-inferiority test; paired design; difference in proportions; comparative use human factors study; Tango method
A comparative use human factors (CUHF) study compares a generic drug-device combination, a complex generic product, to its reference product in terms of the use error rates based on critical tasks impacted by changes in critical external design attributes. The purpose of the CUHF study is to answer the question about whether differences between devices will impact substitutability, i.e. the differences in device usability, by using the non-inferiority test with a pre-specified margin. This is an emerging area with its first draft Guidance published in January 2017. There are several issues related to the design of these studies and analysis of the data from them, including sample size calculation and non-inferiority margin justification. The objective of this project is to examine the commonly used sample size calculation methods and confidence interval estimation methods with the Tango method referenced in the FDA CUHF study draft Guidance 2017. This project considers a model for the difference of two correlated proportions in a matched-pair design of case-control studies. The results from a Monte Carlo simulation study comparing these methods are presented. The study shows that Tango method has empirical significance levels closer to the nominal level and its confidence interval has better empirical coverage probability than the other methods.