Sample size and power analysis for endometrial safety studies

Endometrial safety studies are required for the approval of progestin components. The Committee for Proprietary Medicinal Products requirement is the actual percentage below 2% and the upper limit of the one-sided exact 95% confidence interval not more than 2% above the point estimate. The more recent U.S. Food and Drug Administration requirement is the actual percentage < or = 1% and the upper limit of the one-sided exact 95% confidence interval < or =4%. I studied the sample size and power needed to satisfy both requirements based on the exact confidence intervals for the binomial parameter and the Poisson parameter. I discovered that a larger sample size does not always lead to a higher power. I presented a best sample size that satisfies both requirements and recommended that the patient enrollment should be closely monitored during the study.