Abstract:
|
Bioequivalence trials are on the increase. Sample size estimation procedures for equivalence trials with count data, specifically Poisson endpoints, are a relatively underdeveloped area, with none of the most widely used sample size software incorporating this option. Methods originally developed by Signorini (1991) are used for Poisson regression power calculations, but to date methods testing for equivalence have not been widely developed. Liu (2005) developed an exact calculation to test for equivalence with Poisson data, as well as an approximate sample size calculation formula using the normal approximation, and this approach will be examined. Often in industry, a normal approximation is utilized in order that sample size estimates are provided in reasonable timeframes. Using simulations, an assessment of which specific scenarios the normal approximation is appropriate will be made, and extensions to incorporate a variety of scenarios will be examined including uneven randomization ratios and asymmetric equivalence margins.
|
ASA Meetings Department
732 North Washington Street, Alexandria, VA 22314
(703) 684-1221 • meetings@amstat.org
Copyright © American Statistical Association.