Recurrent Event Analysis Considering Events Duration

• Activity Number: 428
• Type: Contributed
• Date/Time: Tuesday, August 6, 2013 : 2:00 PM to 3:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Kuolung Hu*+
• Companies: Amgen
• Keywords: Recurrent event data ; Anderson-Gill model ; Counting process ; Poisson regression model

Abstract

In some clinical trials, the key endpoint is repeated occurrences of the same or related types of events. Poisson regression and generalized Cox models (Andersen-Gill model and proportional means model (PMM)) can be used in analyzing such data. With the expectation of small subject incidence, how to properly evaluate the treatment effect considering the recurrent event episodes (first and subsequent) is of interest. We assess the operating features of these methods for detecting treatment differences while considering event duration. The Metcalfe and Thompson (2006) approach is used to simulate recurrent events data. Based on our simulations, we found only small discordances in the results from these different analytical approaches. The PMM gave relatively conservative results when the occurrences of subsequent events were correlated.