Sample-Size Calculation for the Andersen-Gill Model in the Presence of Subject Heterogeneity in the Baseline Risk

• Activity Number: 405
• Type: Contributed
• Date/Time: Tuesday, August 3, 2010 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Antje Jahn-Eimermacher*+ and Katharina Ingel
• Companies: University of Mainz and University of Mainz
• Keywords: recurrent events ; survival ; Andersen-Gill

Abstract

In some trials the event of interest can be experienced several times per individual. Unobserved heterogeneity between individuals in the risk for experiencing events causes dependency between events within the same individual. The Andersen-Gill model is often applied for analysis using a robust variance estimate to account for dependencies. These are, however, not considered in the corresponding sample size formula resulting in insufficient power. We investigate how to adjust this formula by a variance inflation factor based on the naive and robust standard error. On simulated data we estimate the variance inflation factor and calculate the empirical power with the sample size formula unadjusted and adjusted by that factor. We show, that the proposed variance inflation factor performs well to determine sample size. We discuss the challenge to estimate that factor prior to study start.