DETERMINING AVERAGE BIOEQUIVALENCE AND CORRESPONDING SAMPLE SIZES IN GENERAL ASYMMETRIC SETTINGS By SAURABH MUKHOPADHYAY

SUMMARY. A new approach has been developed for bioequivalence testing to avoid imprac-tical rejections which may occur with the commonly used Westlake’s (1972) symmetric interval procedure. These methods are shown to cover the original symmetric confidence interval approach as a special case. Also, it is of considerable interest from the manufacturer’s point of view, to ob-tain the minimum sample size which guarantees a pre-specified power to the test. The calculation of this minimum sample size accounts for the plausible asymmetry in the equivalence range and plausible non-zero expected value of the true difference. Algorithms are provided with illustrations for both testing and sample size problems. 1