Applying Weighted GEE for Sample-Size Estimation in Repeated Measurement Studies with Dropout

• Activity Number: 308
• Type: Contributed
• Date/Time: Tuesday, August 6, 2013 : 8:30 AM to 10:20 AM
• Sponsor: Biopharmaceutical Section
• Author(s): Anna Sun*+
• Companies: UMBC
• Keywords: Missing Complete at Random ; Missing at Random ; Missing Data ; Repeated measurements ; Sample size ; Weighted GEE

Abstract

In planning and designing a clinical trial, researchers need to estimate the sample size based on the target power of the statistical test used to test the treatment effect. In studies with repeated measurements, missing data is often inevitable because some patients may drop out before the study ends. Though there are rich literatures in sample size calculation in longitudinal studies, not much work has been done in repeated measurements with missing data, especially missing at random. The available methods either ignore the missing data effect, or assume missingness is MCAR, which is a very strong assumption.

By following weighted GEE theory in Robins, Rotnitzky, and Zhao (1995), we derived a closed form sample size estimation formula for comparing two groups' slope difference for repeated continuous responses under MAR. We also performed simulations to confirm the validity of the weighted GEE method, and showed that generally the power obtained using the WGEE based sample size formula yields a test whose power is very close to the nominal value.