Sample Size and Power for Truncated Bivariate Normal Distribution

• Activity Number: 626
• Type: Contributed
• Date/Time: Thursday, August 5, 2010 : 8:30 AM to 10:20 AM
• Sponsor: Biopharmaceutical Section
• Author(s): Sourav Santra*+
• Companies: Takeda Pharmaceutical Company Limited
• Keywords: sample size ; power ; truncated normal ; censored sample

Abstract

In clinical trials, change from baseline is often used as a primary endpoint. Sample size and power calculations are based on the difference in change from baseline between control and placebo groups under normality assumption. Baseline and final visit observations from a subject is usually highly correlated. Thus the assumption that the data comes from a bivariate normal distribution is reasonable. Further, due to inclusion criteria, baseline observations are truncated (below) and some of the final visit observations may be censored due to limit of detection. Under this more realistic situation, sample size based on conventional method provides less power. Some simulation results will be provided for both superiority and non-inferiority trials to assess the impact of truncation and censoring on statistical power.